Skip to main content
NCBI home page
Food Chemistry: X logoLink to Food Chemistry: X
. 2026 Feb 18;34:103688. doi: 10.1016/j.fochx.2026.103688

Effects of composite steaming and drying process on the taste and aroma of pomelo black tea via metabolomics and sensory histology techniques

Zhiwei Hou a,1, Zhe Gu a,1, Sitong Liu b, Jian Lu c, Huiyuan Zhang a, Le Li a, Yugu Jin a, Hongqing Ye d, Mengqian Lu a, Hongping Chen e,⁎, Huiyan Jia a,⁎
PMCID: PMC12938864  PMID: 41767662

Abstract

This study integrated metabolomics, electronic tongue, and molecular docking to investigate how a multi-stage steam–drying process influences the flavor characteristics of pomelo black tea. Results showed that continuous steam-drying can significantly reduce bitterness and umami, and it could also promote the hydrolysis of flavonoid glycosides such as naringenin and the accumulation of organic acids such as linolenic acid, thereby increasing acidity while reducing bitterness. Volatile component analysis indicated that heat-sensitive terpenes such as γ-terpinene was significantly reduced due to oxidative degradation. Aldehydes and esters, including undecanal and neryl acetate, might compensate for this loss of aroma through thermal conversion. This research also found that hydrogen bonds and hydrophobic interactions were key to the binding of aroma compounds to olfactory receptors. Overall, this study provides molecular- and sensory-level evidence linking moist heat processing to coordinated changes in taste- and aroma-related compounds in pomelo black tea.

Keywords: Pomelo black tea, Metabolomics, Electronic tongue, Flavor quality, Molecular docking

Highlights

  • •

    LC-MS, E-tongue, and GC–MS showed pomelo black tea's flavor and aroma compounds.

  • •

    Composite steam-drying process significantly affected flavor of pomelo black tea.

  • •

    Molecular docking revealed interaction between aroma compound and olfactory receptor.

1. Introduction

Tea is a popular beverage worldwide, and people like drinking it not only for its delightful flavor but also for its many benefits to health (Zhai et al., 2022). Fully fermented black tea accounts for more than 75% of the world's tea consumption (Sharma & Rao, 2009), and has therefore become a focus of further study through processing of the product, owing to its characteristic red-brown color, mellow and sweet taste, and potential roles in weight management and vascular function (Pan et al., 2016; Woodward et al., 2018; Zhang et al., 2019).

Pomelo black tea is a distinctive reprocessed tea formed by integrating fermented black tea with pomelo peel via moist-heat treatment, Owing to its unique citrus–floral aroma and attenuated bitterness, it has garnered growing interest in southern China. Existing studies on black tea flavor formation have primarily focused on single raw material systems and dry-heat processing steps, such as high-temperature drying and roasting that are known to modulate polyphenol degradation, Maillard reactions, and volatile compound formation (Jiang et al., 2022; Wang, Chen, et al., 2024). In parallel, research on citrus or pomelo peel processing has primarily focused on changes in volatile terpenoids during drying, oxidation and storage, typically treating citrus materials as independent systems rather than as components that interact with tea (Nguyen et al., 2009; Yu et al., 2022). However, prior research on tea processing research has seldom investigated how repeated steaming cycles modulate the synergistic transformation of non-volatile flavor compounds and volatile aroma substances, particularly within tea-citrus composite systems.

Pomelo black tea production relies on a composite steaming and baking process, where black tea and pomelo peel are repeatedly exposed to high humidity and high temperature environments. Steaming provides a moist-heat environment that facilitates enzymatic reactions, and promotes the release and interaction of soluble compounds from both tea leaves and pomelo peel. Subsequent drying reduces moisture content and stabilizes intermediate products, thereby fixing the chemical composition formed during steaming. Repetition of this steaming–drying sequence is intended to progressively intensify these transformations. However, quality instability phenomena commonly observed in industrial pomelo black tea production, such as excessive bitterness reduction, aroma imbalance, or freshness loss, remain largely empirical and lack sufficient explanations based on chemical mechanisms. Systematic investigation of the composite steaming and baking process is urgently needed, with particular emphasis on deciphering the cumulative effects of multiple steaming and drying cycles.

Flavor quality is a key determinant of consumer acceptance of pomelo black tea and is governed by the combined perception of taste and aroma. Bitterness, sweetness, and umami are the fundamental descriptive indicators that define the taste of black tea (Wu et al., 2022). These taste attributes result from the combined effects of various flavor compounds, such as catechins, flavonol glycosides, amino acids, caffeine, phenolic acids, theaflavins (TFs), and saccharides (Yu et al., 2025). Meanwhile, the roasting process is very crucial for black tea. Studies show that changes in specific compounds during the roasting process may directly affect the sensory quality of tea (Jiang et al., 2022). However, most of these studies focus on dry-heat treatments or single-material systems, while the effects of repeated moist-heat processing on taste-related metabolites in composite tea products remain poorly understood.

Sensory evaluation remains the predominant method for assessing taste; However, it is inherently constrained by subjectivity, individual preference, and complex compound interactions. As a result, biomimetic sensing technologies, such as the electronic tongue (ET), are gaining increasing recognition. ET is a device designed to mimic human gustatory perception, enabling the analysis and classification of taste profiles in liquid samples. It allows for the rapid quantification of fundamental taste attributes in tea infusions, such as bitterness and umami (Tian et al., 2024). Integrating ET analysis with metabolomics provides an effective strategy for linking sensory characteristics with underlying chemical composition (Ye et al., 2022).

There are more than 600 volatile compounds in black tea, which mainly include aldehydes, alcohols, esters, alkanes, and ketones. The contents of these kinds of volatile substances and individual odor thresholds make up the overall aroma of tea (Kang et al., 2019; Liang et al., 2024). Notably, some volatile compounds can exert a substantial influence on aroma despite their low concentrations, owing to low odor threshold values. Therefore, the Relative Odor Activity Value (ROAV) was proposed to measure the real contribution of these volatiles to the aroma (Wang, Huang, et al., 2024). In addition, molecular docking methods could also be used to investigate the relationship between characteristic aroma substances and olfactory receptors (ORs). This computational approach predicts the interaction between a molecule and its target receptor by evaluating the binding strength and specific types of interactions, such as hydrogen bonding and hydrophobic interactions (Paggi et al., 2024). Olfactory receptors (ORs) are a class of G-protein-coupled receptors that perceive aroma components. The binding of aroma molecules to ORs activates signal transduction pathways, which transmit information to the brain, triggering olfactory perception (Zhu et al., 2024). Molecular docking is a powerful tool for identifying key aroma compounds. It has been widely used in aroma analysis of different kinds of teas (Deng et al., 2024; Liang et al., 2024; Zhu et al., 2024; Zhu et al., 2025).

This study used multi-omics methods to detect chemical compositions. The non-volatile metabolites were analyzed by LC-Orbitrap-MS, and the volatile organic compounds (VOCs) were identified by HS-SPME-GC–MS. Meanwhile, an electronic tongue was used to measure the basic taste attributes of the tea infusion, achieving the mapping of chemical components and sensory characteristics. By systematically evaluating the effects of composite steaming–drying processing on sensory quality, this study aims to provide theoretical guidance for pomelo black tea production and practical insights for the development of tea-based beverages.

2. Materials and methods

2.1. Sample preparation

The black tea samples used in this study were Jingshan black tea, sourced from Zhejiang Hangzhou Jingshan Wufeng Tea Industry Co., Ltd., and made from fresh leaves of the JiuKeng group of tea plants. The Citrus maxima (pomelo) fruits were purchased from the local market in Zhejiang, Hangzhou.

Fresh pomelos with intact peel, uniform shape, thin rind, and pronounced citrus aroma were selected as raw materials. The pomelos were washed thoroughly under running water to remove surface impurities. The pomelos were then opened by cutting a circular lid (approximately 3–5 cm in diameter) at the top, and the pulp was carefully removed to obtain a hollow pomelo peel shell while maintaining its structural integrity.

The black tea was sieved to remove impurities before use. The prepared black tea was filled into the hollow pomelo peel shell and gently compacted to ensure sufficient contact between tea leaves and the inner surface of the peel, while avoiding excessive compression. The removed pomelo lid was then repositioned, and the whole pomelo was fixed using food-grade natural fiber string to maintain shape during subsequent processing.

The black tea and pomelo peel samples were divided into two processing groups: the first group was freeze-dried directly to obtain freeze-dried (FD) samples, while the second group underwent a composite steaming and drying process with the following steps: (1) Primary steaming stage: the black tea was wrapped in peel, placed in a steamer oven at 100 °C for 120 min, and then transferred to a drying box at 120 °C for 120 min to achieve a constant weight, resulting in the first steaming and drying (FSD) samples; (2) Secondary steaming stage: the FSD samples were again subjected to the same steaming (100 °C, 120 min) and drying (120 °C, 120 min) parameters to obtain the secondary steaming and drying (SSD) samples; (3) Tertiary steaming stage: the steaming and drying process was repeated once more to yield the three-stage steaming and drying (TSD) samples. All processed samples (FD, FSD, SSD, TSD) were ground into a homogeneous powder using a pulverizer, mixed in equal mass ratios, sealed, and stored at −4 °C in a calm, light-protected environment for further analysis.

2.2. Reagents and materials

All reagents were of analytical or chromatographic grade. LC-MS grade formic acid, acetic acid, methanol, and acetonitrile were purchased from Thermo Fisher Scientific Inc. (Waltham, MA, USA). Anhydrous ethanol was obtained from Tianjin Damao Chemical Reagent Factory (Tianjin, China). Potassium phosphate monobasic (KH₂PO₄) and silver chloride (AgCl) were supplied by Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). Sodium chloride (NaCl), tartaric acid, and potassium chloride (KCl) were sourced from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). A homologous n-alkane series (C6-C40) for retention index calibration was provided by Sigma-Aldrich (Shanghai, China). Distilled water was produced by Wahaha Group Company (Hangzhou, China). For metabolomic analysis, 4-chloro-DL-phenylalanine was used as the internal standard and was obtained from MedChemExpress (Shanghai, China). Ethyl decanoate, used as the internal standard for volatile compound analysis, was obtained from Shanghai Yuanye Bio-Technology Co., Ltd. (Shanghai, China). The HS-SPME fiber (50/30 μm DVB/CAR/PDMS) and its manual holder were obtained from Supelco (Bellefonte, PA, USA).

2.3. Measurement of taste intensity

An electronic tongue system (TS-5000Z, INSENT, Japan) was employed to evaluate the flavor profile of the tea infusion. The sensors equipped in this system consist of a taste sensor and a ceramic reference electrode. The change between the phase of the sensor film and the reference electrode was transmitted through the potential difference.

All sensors were calibrated and preconditioned prior to measurement. The taste sensors used to measure bitterness, sourness, umami, and sweetness were immersed in a reference solution containing 30 mM KCl and 0.3 mM tartaric acid for adjustment. The ceramic reference electrode was activated for 24 h in a 3.33 M KCl solution containing the same internal electrolyte. Prior to each use, a preliminary test was conducted to verify that the sensor's output voltage was within the standard range and to confirm its proper functioning.

Tea samples were prepared in accordance with GB/T 23776–2018. The ratio of tea to water was 1:50 (g·mL−1). 3.00 g of FD, FSD, SSD, and TSD samples were infused in 150 mL of boiling water for 5 min, filtered through a 400-mesh gauze and cooled to room temperature. 35 mL of the infusion were prepared for electronic tongue measurement. Each infusion type was analyzed in quadruplicate. The first measurement was used to stabilize the sensor, and the average of the remaining three measurements was taken as the final output.

2.4. Analysis of non-volatile metabolites using LC-Orbitrap-MS

The sample extraction method was carried out following the method described in (Zhang et al., 2024). Firstly, 0.4 g of tea powder was combined with 8 mL of 70% methanol solution (containing 200 mg·L−1 4-chloro-DL-phenylalanine as the internal standard) in a 10 mL centrifuge tube. Secondly, The mixture was ultrasonically extracted for 30 min and then allowed to stand for 4 h. After standing, it was centrifuged at 12,000 rpm and 4 °C for 10 min. Finally, a 0.1 mL of the supernatant was transferred to a 5 mL centrifuge tube and diluted with 3.9 mL of 70% methanol (excluding the internal standard). The mixture was placed in a sample bottle for analysis after passing through a 0.22 μm nylon membrane. Quality control (QC) samples were prepared by combining equal volumes of all sample extracts.

Using an ultra-high-performance liquid chromatograph (UltiMate 3000, Dionex, Sunnyvale, CA, USA) coupled with a mass spectrometer (Q-Exactive Focus, Thermo Fisher Scientific, Waltham, MA, USA) to conduct untargeted metabolomics. Chromatographic separation was achieved on an Acquity UPLC HSST3 column (100 mm × 2.1 mm, 1.8 μm, Waters, USA). The specific parameters of the ultra-high-performance liquid chromatograph and mass spectrometer were determined according to a study by(Shen et al., 2022).

2.5. Analysis of volatile metabolites using HS-SPME-GC–MS

The HS-SPME-GC–MS assay followed the method described by (Huang et al., 2022). A 10 mL sample of prepared tea infusion was transferred to a 20 mL headspace sample bottle. Then 3 g of NaCl and 4 μL of ethyl decanoate (10 mg·L-1) were added. The vial was placed in a magnetic rotator to promote the complete evaporation of volatile substances. The sealed vial was incubated in a constant temperature water bath at 50 °C for 15 min. The vial was then inserted into the SPME extraction needle for 40 min. After the extraction, the extraction head was removed and inserted into the GC–MS injection port, where desorption occurred for 5 min at 250 °C.

GC–MS analysis was performed on an Agilent 8890–7000 E instrument (Agilent, USA) with a DB-5MS column (30 m × 0.25 mm × 0.25 μm, Agilent, USA). Separation involved the following temperature program: ramp to 100 °C at 3 °C·min−1, then to 130 °C at 2 °C·min−1, followed by a ramp to 250 °C at 10 °C·min−1, and finally hold at 250 °C for 5 min. High-purity helium (≥ 99.999%) served as the carrier gas. Mass spectra acquisition was performed in positive ion mode, scanning m/z 30–350 with 70 eV electron energy and non-split injection. The MS-MS scan mode was the same as LC-Orbitrap-MS.

2.6. Calculation of ROAV

In this study, the Relative Odor Activity Value (ROAV) was applied to evaluate the contribution of individual volatile compounds to the overall aroma, as determined by GC–MS. The ROAV method reflected each odorant's contribution level to their mixture based on previous research. A high ROAV indicated that a single substance played an important role in defining the flavor characteristic of a sample; compounds with ROAV >1 were usually considered to be key components to define the aroma of tested samples (Xiao et al., 2022).

Determination of ROAV values: OAVi = Ci/OTi, ROAVi = OAVi × (OTmax/Cmax) × 100, where Ci is the relative content of volatile compounds, OTi is the odor threshold of volatile compounds, OTmax is the odor threshold of the volatile compound with the highest OAV, and Cmax is the relative content of volatile compounds with the highest OAV.

2.7. Molecular docking analysis of aroma compound and olfactory receptor interactions

The interaction between olfactory receptors (ORs) and selected aroma compounds was investigated via molecular docking. Five representative broad-spectrum ORs were selected for this study: OR1A1 (UniProt ID: Q9P1Q5), OR1G1 (UniProt ID: P47890), OR2W1 (UniProt ID: Q9Y3N9), OR1D2 (UniProt ID: P34982), and OR52D1 (UniProt ID: Q9H346). These receptors have been reported to interact with various aromatic compounds, including terpenoids, aldehydes, alcohols, and aromatic compounds such as geraniol, citral, decanol, phenylacetone, 2-phenylethanol, and cinnamaldehyde (Sun et al., 2024; Zhu et al., 2025). These compounds were key contributors to the floral, fruity, and citrus aroma profiles commonly observed in tea and citrus products. Owing to their capacity to recognize structurally diverse odorants, these receptors served as ideal vehicles for elucidating the mechanisms underlying interactions with aroma compounds. The protein sequences of the target ORs were obtained from the UniProt database, and their three-dimensional structures were retrieved from the AlphaFold Protein Structure Database, which provides high-accuracy structural predictions based on amino acid sequences. The predicted structures were used as receptor models for subsequent docking analysis. Thirteen key aroma compounds, including nerol, (E)-2-decenal, undecanal, γ-terpinene, nootkatone, methyl salicylate, neryl acetate, α-terpineol, neral, dodecanal, β-thujene, 3-carene, and caryophyllene, were selected as ligands to investigate their interactions with the aforementioned olfactory receptor proteins. The molecular structures of aroma compounds were downloaded from the PubChem database in sdf format and converted to pdb format using OpenBabel. Molecular docking was performed using AutoDockTools and AutoDock (Version 1.5.6, Scripps Research, USA). Prior to docking, receptor proteins and aroma compounds were prepared separately. Water molecules were removed from the receptor structures, polar hydrogen atoms were added, and Gasteiger charges were assigned. Ligand structures were hydrogenated, charged, and rotatable bonds were defined. All prepared receptor and ligand files were saved in pdbqt format.

Potential ligand-binding sites of the receptors were predicted using the DeepSite tool, and the docking grid box was defined to cover the predicted active pocket. The grid box center was set at the predicted binding site, with grid dimensions of 40 Å × 40 Å × 40 Å and a grid spacing of 0.375 Å. A semi-flexible docking strategy was adopted, in which the receptor was kept rigid while the ligands were allowed full conformational flexibility. For each protein–ligand pair, docking simulations were performed to generate multiple binding conformations, and the pose with the lowest predicted binding energy was selected as the optimal binding mode.

The docking results were analyzed to identify binding energies, interaction sites, and key amino acid residues involved in receptor–ligand interactions. Hydrogen bonding and hydrophobic interactions were considered the primary interaction forces. Three-dimensional binding conformations were visualized using PyMOL(Version 2.5.7, DeLano Scientific LLC, USA), and two-dimensional interaction diagrams between aroma compounds and amino acid residues were generated using LigPlot+ (Version 2.2.9).

2.8. Statistical analyses

Raw LC-MS data were converted to mzML format using MSConvert, and the mzML files were subsequently transformed into ABf format with the Analysis Base File Converter. These files were then imported into MS-Dial (version 3.82) for further processing and analyzed in MSFINDER (version 3.04). Identification of metabolites was achieved by comparing MS/MS spectra (Zhang et al., 2024). For statistical analysis, Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA), Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA), and Partial Least Squares Regression (PLSR) were conducted with SIMCA-P (version 14.1, Umetrics, Umea, Sweden). SPSS 27.0.1.0 was used to perform ANOVA, and heat maps were created using TBtools-II (version 2.210).

3. Results and discussion

3.1. Effect of composite steam-drying process on flavor and non-volatile constituents

3.1.1. Quantitative analysis of taste characteristics

The electronic tongue enables objective classification of taste profiles in liquid samples by integrating signals from multiple gustatory-active components (Deng et al., 2022). A systematic evaluation of four key taste attributes (bitterness, sourness, umami, and sweetness) revealed pronounced stage-dependent changes in overall taste profile during processing (Fig. 1) (Table S1). Steaming and drying significantly affected sourness, sweetness, and bitterness (p < 0.01), while umami showed a weaker response (p < 0.05). Bitterness and umami declined from FD to SSD and stabilized at TSD, suggesting that major taste-related transformations plateaued after two steaming–drying cycles. In contrast, sweetness continuously declined at each processing stage, while the sourness increased, indicating that the balance between sourness and sweetness has changed. This change may be associated with the accumulation of organic acids or the loss of alkaline substances.

Fig. 1.

Fig. 1

Open in a new tab

Radar map of taste scores for four samples evaluated using an electronic tongue. “*” indicates significance at p < 0.05 and “**” indicates significance at p < 0.01. FD, the freeze drying sample. FSD, first steaming and drying sample. SSD, second steaming and drying sample. TSD, third steaming and drying sample.

3.1.2. LC-Orbitrap-MS Analysis

To gain a deeper understanding of how steaming and drying affect the metabolism of taste-related compounds, a comprehensive metabolomic analysis was conducted using LC-Orbitrap-MS to characterize chemical transformations throughout pomelo black tea processing.

A total of 1071 metabolite ions were detected in negative mode using LC-Orbitrap-MS metabolomics. Multivariate statistical analysis was performed on the nonvolatile metabolites. PCA of the quality control (QC) samples demonstrated tight clustering (Fig. S1A), indicating high analytical stability. Moreover, PCA score plots revealed distinct spatial separation of samples across the four processing stages, suggesting that the steaming and drying treatments significantly reshaped the metabolite composition. To further visualize inter-stage metabolic differences, OPLS-DA was employed as an exploratory tool (Fig. S1B). OPLS-DA confirmed the model reliability with the validation parameters (R2X = 0.866, R2Y = 0.997, Q2 = 0.979). The results of 200 permutation tests supported the validity of the model, with R2 = 0.701 and Q2 = −0.609, confirming that the model was not overfitted (Fig. S1C). Twenty-nine stage-specific differential metabolites were identified using thresholds of VIP > 1.0 and p < 0.05 (Table S2). These included 11 flavonoids and flavonoid glycosides, 5 organic acids, 3 coumarins, 2 phenolic acids, 2 tannins, 1 catechol, 1 carbohydrate, and 4 other compounds. A heatmap visualization revealed clear stage-dependent patterns in metabolite abundance, underscoring the significant impact of the steaming and drying process on the metabolic landscape of pomelo black tea (Fig. 2A). Notably, dynamic changes in polyphenols, organic acids, and coumarin-related compounds were closely associated with processing progression, suggesting their potential relevance to the sensory characteristics of pomelo black tea.

Fig. 2.

Fig. 2

Open in a new tab

Analysis of non-volatile metabolites and correlation with taste properties. (A) Heat map of non-volatile metabolites of four samples (VIP > 1.0 and p < 0.05); (B) Partial least squares regression (PLSR) analysis of taste and non-volatile metabolites (LC-MS). Non-volatile metabolites was the X variable, ET scores for sweetness, umami, sourness, and bitterness were Y variables. FD, the freeze drying sample. FSD, first steaming and drying sample. SSD, second steaming and drying sample. TSD, third steaming and drying sample.

3.1.3. Flavor modulation by flavonoids

Tea polyphenols were predominantly composed of catechins, flavonoids, and phenolic acids, among which flavonoids served as the primary contributors to the astringency and bitterness of tea infusion (Peluso & Palmery, 2015). Under hot and humid processing conditions, structural modifications and chemical interactions collectively drove dynamic alterations in flavonoid content. The heatmap showed that prunin, naringenin, and luteolin increased progressively across steaming and drying cycle and peaked at the TSD stage (Fig. 2A). This trend might be associated with enzymatic hydrolysis catalyzed by endogenous naringinase in pomelo peel and black tea, which converted naringin into prunin and naringenin. Both compounds exhibited a markedly reduced bitterness, approximately 300-fold lower than their precursor (Seong et al., 2023; Zhu et al., 2017), indicating that their accumulation enhanced the overall flavor profile without amplifying bitterness. Furthermore, luteolin, a naturally occurring antioxidant, might contribute to sensory complexity and oxidative stability.

In contrast, several flavonoid glycosides, such as quercetin-3-galactoside, rutin, kaempferol-3-O-glucoside, and trifolin, decreased from FD to SSD and partially recovered at TSD. Glycosides of kaempferol and quercetin, were known contributors to tea's bitterness and astringency due to their naturally low taste thresholds (0.001–19.80 μmol·L−1) (Wang, Liang, et al., 2024). These were significantly degraded during the initial heat treatments. As such, their breakdown likely led to the observed reduction in the infusion's bitterness. Moreover, glycosidic bond cleavage, either via hydrolase activity or non-enzymatic thermal reactions, might produce aglycones with lower solubility and weaker bitter perception (Wang & Liang et al., 2024). The subsequent partial rebound of glycoside levels at the TSD stage may reflect heat-induced stabilization of intermediates or a shift in secondary metabolic pathways.

3.1.4. Metabolic dynamics of organic and phenolic acids

Organic acids, as essential intermediates in carbohydrate metabolism, significantly influence flavor balance by affecting the tea broth's pH and the solubility of flavor compounds (Qin et al., 2020). In our dataset, linolenic acid, glyceric acid, and erythronic acid increased at the TSD stage and were positively associated with sourness in the PLSR model, suggesting that their accumulation might contribute to the enhanced sour perception. However, these explanations represent only correlational inferences, establishing causality requires targeted experiments. In contrast, malic acid and fumaric acid showed a downward trend throughout the processing stage, indicating their thermal degradation at high temperatures, which was consistent with the previously reported pattern (Xiao, He, et al., 2022). This component reconstruction of polar organic acids might improve the acidity of pomelo black tea. Phenolic acids, important precursors for the biosynthesis of catechins and flavonoids, and contribute to tea flavor and color (Li et al., 2021). Among the detected phenolic acids, the content of protocatechuic acid gradually increased and reached its peak in the TSD stage, while 3-O-p-coumaroylquinic acid had the highest content in the FSD stage. The overall upward trend of these compounds suggest a potential association with flavor transformation, particularly in relation to changes in bitterness, sweetness and acidity.

3.1.5. Flavor transformation of coumarin analogs

As shown in the heatmap (Fig. 2A), the three steaming and drying processes caused a significant decrease in the levels of two coumarin compounds, methoxsalen and columbianetin acetate, in the pomelo black tea. Coumarins, a class of secondary metabolites in plants, are typically associated with herbaceous or woody aromatic notes and may impart slight bitterness or sweetness to foods. Their characteristic furanopyranone skeleton is thermally labile, rendering them susceptible to oxidative degradation or hydrolysis under elevated temperature and humidity. Steaming and drying processes was associated with a gradual reduction in the contents of methoxsalen, which induced bitter almond flavor. In addition, columbianetin acetate might be hydrolyzed to acetic acid and columbianetin under high temperature and humidity. Acetic acid may interact with terpenoids such as limonene, potentially contributing to sourness perception. The sensory profile of pomelo black tea was driven by the continual remodeling of its chemical constituents, including flavonoids, organic acids, and coumarin analogs, during complicated steaming and drying processes. The decomposition of glycosides concurrent with the buildup of organic acids might contribute to diminishing bitterness, enhancing sourness, and achieving a more harmonious flavor equilibrium.

3.1.6. Correlation analysis of flavor-presenting compounds

To quantitatively examine the relationship between the metabolite and flavor in pomelo black tea, partial least squares regression (PLSR) analysis was applied. The PLSR analysis revealed correlation between non-volatile metabolites and taste attributes (Fig. 2B). Metabolite signal intensities were used as the independent variables (X). The sensory taste values, including bitterness, sweetness, umami, and sourness, were treated as the dependent variables (Y). Variables within the same quadrant showed a positive correlation, and the greater the distance from the origin, the greater the influence (Zhang et al., 2024). According to the PLSR loading diagram, four flavonoids and flavonoid glycosides: trifolin (ID: 17701), kaempferol-3-O-glucoside (ID: 17704), quercetin 3-galactoside (ID: 18443), and rutin (ID: 24171) showed positive associations with bitterness. These findings were consistent with previous studies that flavonoid glycosides contribute to astringency and bitterness of tea infusions due to their low taste threshold (Zhang et al., 2020).

In addition, the coumarins methoxsalen (ID: 5915) and columbianetin acetate (ID: 9560) were positively associated with umami, sweetness, and bitterness, but negatively associated with sourness. This supported earlier findings that decreases in these compounds during processing coincided with reductions in umami and sweetness, alongside elevated acidity, corroborating electronic tongue results. Notably, the organic acids glyceric acid (ID: 242), erythronic acid (ID: 1722), and linolenic acid (ID: 9068), as well as the phenolic acid 3-O-p-coumaroylquinic acid (ID: 12172), showed strong positive correlations with sourness, indicating their key roles in acidity perception.

These results suggested that the composite steam-drying process was associated with sensory properties through coordinated changes in multiple metabolite classes. Specifically, the attenuation of bitterness, enhancement of sourness, and overall flavor complexity might be related to multi-pathway transformation of flavonoids, coumarins, and organic acids.

3.2. Effect of compound steam-drying process on aroma and volatile components

3.2.1. Analysis of total and differential volatile compounds

To investigate the impact of the composite steaming-drying process on the aroma profile of pomelo black tea, headspace solid phase microextraction (SPME) coupled with gas chromatography–mass spectrometry (GC–MS) were employed. Across four processing stages (FD, FSD, SSD, and TSD), 143 volatile compounds were identified through the NIST chemistry database and retention index comparison (Table S3). The distribution of volatile compounds varied significantly across stages, with 84, 89, 74, and 76 compounds detected in the FD, FSD, SSD, and TSD samples, respectively, indicating substantial phase-specific volatility profiles.

As illustrated in Fig. 3A, the PCA score plot displayed a clear separation of the samples, indicating that the steaming and drying process markedly altered the volatile composition. ANOVA combined with Duncan's multiple range test (p < 0.05) identified 29 differential volatiles with significant stage-dependent variation (Table S4). Hierarchical clustering analysis demonstrated that the FSD samples clustered independently, highlighting a unique intermediate volatile profile formed during the first steaming and drying stage (Fig. 3B). As depicted in Fig. 3C, the compositional shifts in volatile classes were also stage-dependent. The relative proportion of terpenes consistently increased with the number of processing cycles, while alcohols showed a decreasing trend. This inverse relationship suggested that the thermal and oxidative conditions during processing may favor terpene accumulation and alcohol reduction.

Fig. 3.

Fig. 3

Open in a new tab

Analysis of differential volatile compounds of four samples. (A) Principal Component Analysis (PCA); (B) Hierarchical Cluster Analysis (HCA); (C) Percentage chart of volatile compounds; (D) Venn diagrams. FD, the freeze drying sample. FSD, first steaming and drying sample. SSD, second steaming and drying sample. TSD, third steaming and drying sample.

Venn diagram analysis revealed that the number of stage-specific volatile compounds was highest in the FSD group, with 29 unique components, compared to 19 in FD, 5 in SSD, and 7 in TSD. Only 37 compounds (25.7%) were common across all stages, reinforcing the strong remodeling effect of thermal processing on the volatile landscape (Fig. 3D). Unique compounds in the FSD stage might reflect intermediate-phase transformations such as Maillard reactions, ester hydrolysis, and new aroma compound synthesis under moist heat exposure. These results suggested that the first steaming and drying phase was pivotal in restructuring the volatile profile and generating key aroma precursors.

3.2.2. Screening of aromatic active compounds

Although multivariate statistical analyses identified volatile compounds exhibiting significant variation across the steaming and drying stages, their sensory contributions remained uncertain owing to differences in odor thresholds. To address this, the Relative Odor Activity Value (ROAV) method was applied to quantify the contribution of each compound to the overall aroma. Based on ROAV >1 and p < 0.05, thirteen key aroma-active compounds were identified from the four pomelo black tea samples (FD, FSD, SSD, TSD): nerol, (E)-2-decenal, undecanal, γ-terpinene, nootkatone, methyl salicylate, neryl acetate, α-terpineol, neral, dodecanal, β-thujene, 3-carene, caryophyllene. As summarized in Table 1, these compounds interacted synergistically to form the complex aroma profile of pomelo black tea. The dynamic changes in their ROAVs was illustrated in Fig. 4, highlighting key aroma transitions across processing stages.

Table 1.

Identification of key metabolites in samples through GC–MS (ROAV >1.0 and p < 0.05).

No. CAS Caompound Aroma descriptiona OTs (μg/kg)b Concentration (μg/kg)c
 ROAV
FDd FSD SSD TSD FD FSD SSD TSD
1 98–55-5 α-Terpineol floral, citrus 330 15.52 ± 7.63a 1.27 ± 0.10b 1.34 ± 0.51b 1.02 ± 0.22b 0.67 0.28 1.11 1.26
2 106–25-2 Nerol rose-like, floral 300 34.73 ± 12.88a 6.88 ± 0.24b 4.92 ± 0.61b 2.86 ± 0.92b 1.64 1.66 4.52 3.89
3 3913-81-3 (E)-2-Decenal green, fatty 0.3 0.22 ± 0.08a 0.06 ± 0.01b 0.10 ± 0.08ab n.d. 10.24 14.80 89.26 n.d.
4 106–26-3 Neral citral, lemon 100 39.87 ± 24.01a 0.71 ± 0.12b 2.84 ± 2.91b 0.87 ± 0.12b 5.65 0.51 7.82 3.55
5 112–44-7 Undecanal floral, citrus 5 0.13 ± 0.05b 0.14 ± 0.01ab 0.22 ± 0.10ab 0.26 ± 0.04a 1.18 9.68 51.70 100
6 112–54-9 Dodecanal floral, citrus 2 0.41 ± 0.22c 0.67 ± 0.02bc 0.94 ± 0.28ab 1.22 ± 0.25a 0.89 5.20 29.89 52.60
7 28,634–89-1 β-Thujene – 980 27.25 ± 19.50a 1.43 ± 0.41b 3.33 ± 0.61b 4.05 ± 1.51b 0.39 0.11 0.94 1.69
8 13,466–78-9 3-Carene pine, citrus 770 19.10 ± 11.65ab 9.58 ± 2.77b 19.65 ± 3.88ab 30.18 ± 12.92a 0.35 0.90 7.03 16.02
9 99–85-4 γ-Terpinene lemon, herbal 0.26 1.83 ± 1.02a 0.36 ± 0.11b 0.09 ± 0.08b n.d. 100 100 100 n.d.
10 87–44-5 Caryophyllene wood, spice 64 0.21 ± 0.25b 4.23 ± 0.50a 0.31 ± 0.07b 0.46 ± 0.14b 0.05 4.77 1.34 2.91
11 4674-50-4 Nootkatone grapefruit-like 0.8 0.33 ± 0.12a 0.06 ± 0.02b 0.06 ± 0.04b 0.02 ± 0.00b 5.92 5.64 20.40 12.53
12 119–36-8 Methyl salicylate mint-like 40 23.84 ± 8.71a n.d. 1.08 ± 0.20b 0.45 ± 0.11b 8.45 n.d. 7.44 4.58
13 141–12-8 Neryl acetate floral, citrus 42 6.30 ± 2.77a 1.14 ± 1.60b 5.74 ± 3.44ab 5.45 ± 0.99ab 2.13 1.97 37.63 53.04
Open in a new tab

Note: a aroma description of compounds referred to the website: Perflavory Search (http://www.perflavory.com/search.php) (“-”, Indicates not retrieved). b Odor thresholds (OTs) in water from databases and literature. All odor thresholds were obtained from: Leffingwell & Associates (http://www.leffingwell.com/odorthre.htm), (Cheng et al., 2024), (Ye et al., 2024), (TAMURA et al., 2001). c Mean values of triplicates with standard deviations (SDs). The freeze drying (FD), first steaming and drying (FSD), second steaming and drying (SSD), and third steaming and drying (TSD) samples (“n.d.”, the compounds were not detected). d The lowercase letters (a-c) following the values in the different sample entries in the same row are significantly different according to the Duncan test (p < 0.05).

Fig. 4.

Fig. 4

Open in a new tab

The concentration of key metabolites in four samples (ROAV >1.0, p < 0.05). FSD, first steaming and drying sample. SSD, second steaming and drying sample. TSD, third steaming and drying sample. Aroma description: γ-Terpinene (lemon odor), Neryl acetate (floral and citrus odor), Dodecanal (floral and citrus odor), Undecanal (floral and citrus odor), Nootkatone (grapefruit-like odor), 3-Carene (citrus odor), Neral (lemon and citrus odor), Nerol (floral odor), α-Terpineol (floral and citrus odor), (E)-2-Decenal (green odor), Methyl salicylate (mint-like odor), Caryophyllene (spice odor), β-Thujene. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Among these, γ-terpinene emerged as the dominant aroma-active contributor (ROAV = 100) across the FD, FSD, and SSD stages. This consistent with its well-documented association with lemon and citrus aromas in tea products (Guo et al., 2019). However, its concentration declined progressively and became undetectable at the TSD stage. This reduction was likely due to its thermal lability, as γ-Terpinene was a monoterpene prone to oxidative degradation under moist heat, accelerated by elevated temperature and humidity (Nguyen et al., 2009).

Other terpenoids, such as α-terpineol, nerol, nootkatone, and neral, also exhibited similar thermal sensitivity, which was consistent with previous research results that their unsaturated hydrocarbon skeletons were susceptible to oxidation and decomposition under wet heat stress (Kfoury et al., 2017; Tan et al., 2018). These compounds were major contributors to pomelo and citrus aromas (Chaudhary et al., 2018; Duan et al., 2025), so their decrement would probably be responsible for reduced pomelo-like aromas at the late period of processing stages.

Contents of methyl salicylate decreased successively among FSD, SSD, and TSD stages (p < 0.05). One possible explanation was that prolonged exposure to a high-temperature environment might suppress phenylalanine pathway-mediated biosynthesis (Ma et al., 2018; Wang, Qin, et al., 2023). This compound added a refreshing minty characteristic to the tea infusion (Wang, Yu, et al., 2023), and the decrease of methyl salicylate might help in the transformation from fresh to floral and citrus flavors in pomelo black tea.

During TSD stage, aldehydes and monoterpene compounds such as undecanal, dodecanal, and 3-carene increased significantly (p < 0.05). These compounds enhanced the citrus floral aroma and sustained fragrance intensity. The ROAV of neryl acetate, a key floral-scented ester, showed a significant increase, particularly at the SSD and TSD stages. This indicated that heat treatment might activate the esterification reaction.

3.3. Molecular docking

3.3.1. Key interaction regions of OR and aroma compound complexes

Olfactory receptors (OR) were membrane proteins that can detect volatile ligands and initiate intracellular signaling in response to odor stimuli (Sun et al., 2024). In this study, five representative human ORs (OR1D2, OR1G1, OR1A1, OR2W1, and OR52D1) were selected to investigate binding interactions with key aroma compounds from pomelo black tea. These receptors, ranging from 309 to 320 amino acids in length, shared a conserved structure with an extracellular N-terminus, an intracellular C-terminus, seven transmembrane helices (TM1-TM7), three extracellular loops (EC1-EC3), and three intracellular loops (IC1-IC3) (Xiao et al., 2024).

The functional diversity of ORs was largely associated with variations in amino acid sequences, which influence ligand-binding specificity. Molecular docking analysis indicated that different aromatic compounds selectively bind to specific residues in different receptor protein regions. As shown in Table S5, characteristic interactions were observed at: Asn155 (TM4, OR1A1), Leu14 (EC1, OR1G1), Thr240 (TM6, OR2W1), Ser290 (TM7, OR1D2), and Ser116 (TM3, OR52D1). Differences in these binding sites and residues underlie variations in ligand binding affinity.

3.3.2. Relative binding strengths across different ORs

Binding energy was a key metric for evaluating the interaction strength between aroma compounds and olfactory receptors (ORs). Lower binding energy values reflected stronger ligand-receptor affinities, implying a higher likelihood of receptor activation. Upon binding, ORs initiated intracellular signaling cascades that contributed to aroma perception. Therefore, comparative binding energy analysis was used to provide structural insight into the potential involvement of these compounds in aroma perception.

As presented in Table 2, binding energies between the thirteen key aroma compounds and five ORs ranged from −8.75 to −2.62 kcal/mol, suggesting thermodynamically favorable and spontaneous interactions. Importantly, the wide binding energy range also indicated substantial variability in interaction strength among different OR subtypes, suggesting that each receptor exhibits distinct sensitivity toward individual aroma compounds. This variability underscores the ligand discrimination specificity of each receptor subtype. Such receptor-dependent variability was consistent with the combinatorial receptor code, where one odorant could be recognized by multiple ORs with different sensitivities and interaction features (Malnic et al., 1999). The average binding energies of the tested compounds across the five receptors were as follows (Fig. 5, Fig. S2–13): nerol (−4.29 kcal/mol), (E)-2-decenal (−4.01 kcal/mol), undecanal (−3.95 kcal/mol), γ-terpinene (−5.43 kcal/mol), nootkatone (−7.31 kcal/mol), methyl salicylate (−4.13 kcal/mol), neryl acetate (−3.90 kcal/mol), α-terpineol (−6.41 kcal/mol), neral (−3.98 kcal/mol), dodecanal (−3.87 kcal/mol), β-thujene (−5.30 kcal/mol), 3-carene (−5.79 kcal/mol), and caryophyllene (−8.00 kcal/mol). Crucially, lower binding energy correlated with stronger compound-receptor affinity. This results suggested that these aroma compounds preferentially recognize and activate receptors to elicit characteristic aromas. For instance, OR1D2 preferentially interacted with terpenoid hydrocarbons and sesquiterpenes exhibiting lower binding energies, whereas OR1G1 and OR52D1 showed stronger affinities toward oxygenated compounds such as nootkatone and α-terpineol. These patterns suggest functional differentiation among OR subtypes in recognizing specific chemical classes of aroma compounds. This analysis indicated that compounds such as caryophyllene, nootkatone, and α-terpineol showed relatively lower predicted binding energies compared with other aroma compounds. Notably, these compounds were major contributors to spice-like, grapefruit-like, floral and citrus aromas.

Table 2.

Summary of the binding energy between olfactory receptors and ligands.

Liganda Binding energy (kcal/mol)c
OR1A1b OR1G1 OR2W1 OR1D2 OR52D1
Nerol −4.96 −4.39 −4.53 −4.19 −3.38
(E)-2-Decenal −5.04 −3.56 −4.37 −3.74 −3.32
Undecanal −4.36 −4.37 −3.85 −4.01 −3.14
γ-Terpinene −5.92 −5.81 −5.37 −5.06 −5.00
Nootkatone −8.75 −8.63 −6.25 −6.61 −6.29
Methyl salicylate −4.55 −5.4 −3.58 −4.25 −2.89
Neryl acetate −5.17 −3.7 −3.83 −3.99 −2.79
α-Terpineol −7.04 −7.21 −6.25 −5.77 −5.78
Neral −4.23 −4.57 −4.47 −4.00 −2.62
Dodecanal −3.93 −2.94 −4.71 −3.98 −3.81
β-Thujene −5.67 −5.64 −5.33 −5.09 −4.79
3-Carene −6.13 −5.99 −5.77 −5.5 −5.55
Caryophyllene −8.42 −8.54 −7.97 −7.60 −7.48
Open in a new tab

Note: a Thirteen important compounds in pomelo black tea were selected as ligands. b Five proteins were selected as the olfactory receptors. c The binding energy between olfactory receptors and ligands.

Fig. 5.

Fig. 5

Open in a new tab

Molecular docking simulation between nootkatone and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1.

Notably, methyl salicylate showed the greatest relative difference in binding energy across receptors, with a 46.48% variation between its interactions with OR1G1 and OR52D1, emphasizing significant receptor-ligand selectivity. Likewise, nerol bound to OR1D2, OR1G1, OR1A1, OR2W1, and OR52D1 in various structural regions including EC3, IC1, EC3, TM3, and TM7, interacting with residues such as Tyr182, Leu55, Asp180, Asn155, and Ala281 respectively. Differences in interaction sites and involved amino acid residues might contribute to variations in binding affinity and selectivity, which was consistent with the complexity of olfactory recognition in practical aroma perception. Moreover, individual ORs exhibited heterogeneous responses to different compounds. For example, binding energies between OR1D2 and the thirteen aroma compounds spanned from −3.74 to −7.60 kcal/mol, representing a 50.79% energy difference. Similar maximum energy variation rates were observed for OR1G1 (65.93%), OR1A1 (55.09%), OR2W1 (55.08%), and OR52D1 (64.97%), confirming a high degree of receptor-specific binding selectivity.

From a mechanistic perspective, variations in relative binding strength across OR subtypes might influence the magnitude and combination of downstream signaling events, thereby shaping the complexity and layering of perceived aromas. Collectively, the heterogeneous binding patterns across OR subtypes suggested that key volatiles might differentially engage multiple receptors, which could influence the overall receptor activation pattern underlying practical aroma perception (Buck, 2004). Rather than being driven by a single compound-receptor pair, the aroma profile of pomelo black tea was more likely shaped by the collective engagement of multiple aroma-active compounds with multiple ORs. In this context, differences in ORs and receptor binding sites might influence how aroma signals were integrated and perceived, thereby contributing to the characteristic and layered aroma attributes of pomelo black tea.

In this context, binding energy values represented predicted interaction affinity under static conditions and do not directly correspond to aroma intensity or sensory strength. Therefore, these docking results provided a possible structural basis for their sensory relevance. Such binding potential represented a necessary but not sufficient condition for actual aroma perception. Molecular docking model does not account for critical factors such as the compound's concentration in the headspace, its odor threshold, or its contribution to the overall volatile profile of the tea sample. This interpretation was consistent with previous research in which molecular docking has been used to better understand odor perception and to provide insights into the mechanism underlying characteristic aroma formation (Sun et al., 2024; Zhu et al., 2025). Even if a direct quantitative correlation with sensory intensity cannot be established from the present data, but offered a mechanistic hypothesis, they established a molecular theoretical foundation for future research to establish direct quantitative correlations between specific aroma compound-receptor interactions and sensory intensity.

3.3.3. Analyzing the type of forces between OR and aroma compounds

The interaction sites between thirteen representative aroma compounds and five olfactory receptors (ORs) were summarized in Table S5, revealing the diversity of binding mechanisms. For example, neryl acetate bound to Thr182 in OR1A1, nerol to Leu55 in OR1G1, undecanal to Asp175 in OR2W1, (E)-2-decenal to Tyr155 in OR1D2, and dodecanal to His184 in OR52D1. Binding interactions revealed that hydrogen bonding was a potential mechanism for most compounds, with only γ-terpinene and β-thujene failing to interact with the studied receptors in this manner. These hydrogen bonds primarily involved polar residues located within transmembrane helices or extracellular loop regions, where functional groups such as hydroxyl, carbonyl, or ester moieties of aroma compounds acted as hydrogen bond donors or acceptors, thereby stabilizing ligand positioning within the receptor binding pocket. Notably, nootkatone established a hydrogen bonding network with two residues, Asn84 and Leu14, within OR1G1 (Fig. 5), with a relatively low binding energy of −8.63 kcal/mol. The formation of multiple hydrogen bonds suggests a cooperative interaction mechanism, in which simultaneous polar contacts enhance ligand stabilization and binding affinity. Polar functional groups in nootkatone, particularly the ketone and alkene moieties, promote interactions with electron-donating or electron-accepting side chains such as the amide group in Asn84. These findings underscored hydrogen bonding as a critical determinant of OR-ligand interactions.

Remarkably, despite not forming any hydrogen bonds, γ-terpinene and β-thujene still exhibited strong binding affinities, with average binding energies of −5.43 kcal/mol and − 5.30 kcal/mol, respectively. Both values were lower than that of nerol (−4.29 kcal/mol), which did form hydrogen bonds with all five ORs. This suggested that non-polar forces, particularly hydrophobic interactions, also dominate the OR recognition of specific ligands. Analysis of these hydrophobic interactions revealed that γ-terpinene interacted with seven hydrophobic residues (Leu44, Ser290, Leu300, Leu291, Phe287, Pro286, Gly41) within OR1D2. The proximity of these residues formed a continuous hydrophobic environment that stabilized the γ-terpinene molecule. This structural context was essential for maintaining the integrity and perception of its characteristic lemon and citrus aroma. Previous research confirmed that hydrophobic residues facilitated stable ligand binding by creating an energetically favorable microenvironment (Zeng et al., 2023).

Similarly, even though nootkatone couldn't form hydrogen bonds with OR52D1, it formed favorable interactions with six hydrophobic residues: Leu119, Ala120, Ser116, Gly151, Ile147, Phe123, and the binding energy of nootkatone was −6.29 kcal/mol, lower than that of neral interacting with His141 (−2.62 kcal/mol) This structure, formed by aliphatic chains and aromatic residues such as Ile147 and Phe123, helped stabilize the isopropenyl skeleton of nootkatone, which might be responsible for the release of grapefruit-like aroma. These results suggested that, besides the formation of hydrogen bonds between ligand and ORs, hydrophobic interactions were also the basis of ligand interaction and activation of the olfactory receptor, especially in terms of volatile substances with non-polar characteristics (Sun et al., 2024).

4. Conclusion

This study employed a multi-omics approach to comprehensively characterize the coordinated alterations in chemical composition and sensory properties of pomelo black tea resulting from a composite steaming-drying process. Significant correlations were observed between specific metabolite classes (flavonoid glycosides, coumarins, and organic acids) and distinct sensory taste attributes. Under hot and humid processing conditions, multi-stage steaming and drying promoting the hydrolysis of bitter flavonoid glycosides reducing perceived bitterness. Concurrently, the accumulation of organic acids was associated with heightened acidity, which further contributed to the accumulation of acidity. Volatile compound analysis identified thirteen key aroma-active compounds associated with pomelo, citrus, minty, and floral aromas in the processed samples, revealing their dynamic evolution across thermal processing stages and elucidating their respective contributions to the distinctive aromatic profile of each phase. Furthermore, molecular docking analyses of these key aroma compounds with five representative olfactory receptors (OR1D2, OR1G1, OR1A1, OR2W1, and OR52D1) revealed distinct binding patterns, highlighting the critical contributions of hydrogen bonding and hydrophobic interactions to ligand–receptor recognition. Collectively, these findings provide comprehensive sensory, metabolic, and molecular evidence directly linking composite steam–drying processing to the systematic flavor chemistry of pomelo black tea, thereby establishing a scientific foundation for process optimization tailored to flavor quality.

CRediT authorship contribution statement

Zhiwei Hou: Writing – original draft, Software, Methodology, Funding acquisition, Formal analysis, Data curation, Conceptualization. Zhe Gu: Writing – original draft, Software, Methodology, Funding acquisition, Formal analysis, Data curation. Sitong Liu: Methodology, Investigation, Data curation. Jian Lu: Software, Methodology, Formal analysis. Huiyuan Zhang: Software, Methodology, Formal analysis. Le Li: Methodology, Investigation, Data curation. Yugu Jin: Methodology, Investigation, Data curation. Hongqing Ye: Visualization, Supervision, Investigation. Mengqian Lu: Visualization, Supervision, Investigation. Hongping Chen: Resources. Huiyan Jia: Writing – review & editing, Visualization, Supervision, Investigation.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors acknowledge the financial support of National Natural Science Foundation of China (32302608), The Open Fund of National Key Laboratory for Tea Plant Germplasm Innovation and Resource Utilization (NKLTOF20240115), the Scientific Research and Development Foundation of Zhejiang A & F University (2022FR025), Zhejiang University Student Science and Technology Innovation Activity Plan (New Seedling talent Plan subsidy project, 2025R412B042).

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.103688.

Contributor Information

Hongping Chen, Email: thean27@tricaas.com.

Huiyan Jia, Email: jhy250123@163.com.

Appendix A. Supplementary data

Supplementary material: Figure S1 Analysis of differential non-volatile metabolites of four samples. (A) Principal Component Analysis (PCA); (B) Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA); (C) results of 200 permutation test. FD, the freeze drying sample. FSD, first steaming and drying sample. SSD, second steaming and drying sample. TSD, third steaming and drying sample. Figure S2 Molecular docking simulation between 3-carene and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S3 Molecular docking simulation between caryophyllene and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S4 Molecular docking simulation between (E)-2-decenal and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S5 Molecular docking simulation between dodecanal and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S6 Molecular docking simulation between methyl salicylate and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S7 Molecular docking simulation between neral and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S8 Molecular docking simulation between nerol and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S9 Molecular docking simulation between neryl acetate and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S10 Molecular docking simulation between γ-terpinene and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S11 Molecular docking simulation between undecanal and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S12 Molecular docking simulation between α-terpineol and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S13 Molecular docking simulation between β-thujene and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Table S1 Electronic tongue test results of pomelo black tea. Table S2 Identification of key metabolites in samples through LC-MS (VIP > 1.0 and p < 0.05). Table S3 Total volatile compounds in four samples. Table S4 Volatile compounds with p < 0.05 in four samples. Table S5 Summary of the interaction forces between olfactory receptors and ligands.

mmc1.docx (3MB, docx)

Data availability

Data will be made available on request.

References

  1. Buck L.B. Olfactory receptors and Odor coding in mammals. Nutrition Reviews. 2004;62(s3):S184–S188. doi: 10.1111/j.1753-4887.2004.tb00097.x. [DOI] [PubMed] [Google Scholar]
  2. Chaudhary, P. R., Jayaprakasha, G. K., & And Patil, B. S. (2018). Identification of volatile profiles of Rio red grapefruit at various developmental to maturity stages. Journal of Essential Oil Research, 30(2), 77–83. Doi: / 10.1080/10412905.2017.1386131. [DOI]
  3. Deng G., Huang L., Wang W., Yu T., Ning J., Wang Y. Mechanism of key volatiles in osmanthus scented green tea on the improvement of its aroma and taste quality. Food Chemistry. 2024;460 doi: 10.1016/j.foodchem.2024.140560. [DOI] [PubMed] [Google Scholar]
  4. Deng S., Zhang G., Olayemi Aluko O., Mo Z., Mao J., Zhang H.…Liu H. Bitter and astringent substances in green tea: Composition, human perception mechanisms, evaluation methods and factors influencing their formation. Food Research International. 2022;157 doi: 10.1016/j.foodres.2022.111262. [DOI] [PubMed] [Google Scholar]
  5. Duan Y., Liu Z., Zhang Q., Hu J., Zheng S., Mao X., Wang Y. Dynamic changes in volatile and non-volatile components of Jinggang honey pomelo (Citrus maxima (L.) Osbeck) during growth revealed through an untargeted metabolomics approach. Food Chemistry. 2025;478 doi: 10.1016/j.foodchem.2025.143654. [DOI] [PubMed] [Google Scholar]
  6. Guo X., Ho C., Schwab W., Song C., Wan X. Aroma compositions of large-leaf yellow tea and potential effect of theanine on volatile formation in tea. Food Chemistry. 2019;280:73–82. doi: 10.1016/j.foodchem.2018.12.066. [DOI] [PubMed] [Google Scholar]
  7. Huang W., Fang S., Wang J., Zhuo C., Luo Y., Yu Y., Li L., Wang Y., Deng W., Ning J. Sensomics analysis of the effect of the withering method on the aroma components of Keemun black tea. Food Chemistry. 2022;395 doi: 10.1016/j.foodchem.2022.133549. [DOI] [PubMed] [Google Scholar]
  8. Jiang Z., Han Z., Wen M., Ho C., Wu Y., Wang Y., Xu N., Xie Z., Zhang J., Zhang L., Wan X. Comprehensive comparison on the chemical metabolites and taste evaluation of tea after roasting using untargeted and pseudotargeted metabolomics. Food Science and Human Wellness. 2022;11(3):606–617. doi: 10.1016/j.fshw.2021.12.017. [DOI] [Google Scholar]
  9. Kang S., Yan H., Zhu Y., Liu X., Lv H., Zhang Y.…Lin Z. Identification and quantification of key odorants in the world’s four most famous black teas. Food Research International. 2019;121:73–83. doi: 10.1016/j.foodres.2019.03.009. [DOI] [PubMed] [Google Scholar]
  10. Kfoury M., Landy D., Ruellan S., Auezova L., Greige-Gerges H., Fourmentin S. Nootkatone encapsulation by cyclodextrins: Effect on water solubility and photostability. Food Chemistry. 2017;236:41–48. doi: 10.1016/j.foodchem.2016.12.086. [DOI] [PubMed] [Google Scholar]
  11. Li Q., Jin Y., Jiang R., Xu Y., Zhang Y., Luo Y., Huang J., Wang K., Liu Z. Dynamic changes in the metabolite profile and taste characteristics of Fu brick tea during the manufacturing process. Food Chemistry. 2021;344 doi: 10.1016/j.foodchem.2020.128576. [DOI] [PubMed] [Google Scholar]
  12. Liang Y., Wang Z., Zhang L., Dai H., Wu W., Zheng Z., Lin F., Xu J., Huang Y., Sun W. Characterization of volatile compounds and identification of key aroma compounds in different aroma types of Rougui Wuyi rock tea. Food Chemistry. 2024;455 doi: 10.1016/j.foodchem.2024.139931. [DOI] [PubMed] [Google Scholar]
  13. Ma C., Li J., Chen W., Wang W., Qi D., Pang S., Miao A. Study of the aroma formation and transformation during the manufacturing process of oolong tea by solid-phase micro-extraction and gas chromatography–mass spectrometry combined with chemometrics. Food Research International. 2018;108:413–422. doi: 10.1016/j.foodres.2018.03.052. [DOI] [PubMed] [Google Scholar]
  14. Malnic B., Hirono J., Sato T., Buck L.B. Combinatorial receptor codes for Odors. Cell. 1999;96(5):713–723. doi: 10.1016/S0092-8674(00)80581-4. [DOI] [PubMed] [Google Scholar]
  15. Nguyen H., Campi E.M., Roy Jackson W., Patti A.F. Effect of oxidative deterioration on flavour and aroma components of lemon oil. Food Chemistry. 2009;112(2):388–393. doi: 10.1016/j.foodchem.2008.05.090. [DOI] [Google Scholar]
  16. Paggi J.M., Pandit A., Dror R.O. The art and science of molecular docking. Annual Review of Biochemistry. 2024;93:389–410. doi: 10.1146/annurev-biochem-030222-120000. Volume 93, 2024. [DOI] [PubMed] [Google Scholar]
  17. Pan H., Gao Y., Tu Y. 2016. Mechanisms of Body Weight Reduction by Black Tea Polyphenols Molecules (21, pp.) [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Peluso I., Palmery M. Flavonoids at the pharma-nutrition interface: Is a therapeutic index in demand? Biomedicine & Pharmacotherapy. 2015;71:102–107. doi: 10.1016/j.biopha.2015.02.028. [journal article; review] [DOI] [PubMed] [Google Scholar]
  19. Qin D., Wang Q., Li H., Jiang X., Fang K., Wang Q.…Wu H. Identification of key metabolites based on non-targeted metabolomics and chemometrics analyses provides insights into bitterness in Kucha [Camellia kucha (Chang et Wang) Chang] Food Research International. 2020;138 doi: 10.1016/j.foodres.2020.109789. [DOI] [PubMed] [Google Scholar]
  20. Seong H., Im A.E., Kim H., Park N., Yang K., Kim D., Nam S. Production of prunin and naringenin by using Naringinase from aspergillus oryzae NYO-2 and their neuroprotective properties and Debitterization. Journal of Agricultural and Food Chemistry. 2023;71(3):1655–1666. doi: 10.1021/acs.jafc.2c06586. [DOI] [PubMed] [Google Scholar]
  21. Sharma V., Rao L.J.M. A thought on the biological activities of black tea. Critical Reviews in Food Science and Nutrition. 2009;49(5):379–404. doi: 10.1080/10408390802068066. [DOI] [PubMed] [Google Scholar]
  22. Shen S., Huang J., Li T., Wei Y., Xu S., Wang Y., Ning J. Untargeted and targeted metabolomics reveals potential marker compounds of an tea during storage. LWT- Food Science and Technology. 2022;154 doi: 10.1016/j.lwt.2021.112791. [DOI] [Google Scholar]
  23. Sun Z., Lin Y., Yang H., Zhao R., Zhu J., Wang F. Characterization of honey-like characteristic aroma compounds in Zunyi black tea and their molecular mechanisms of interaction with olfactory receptors using molecular docking. LWT- Food Science and Technology. 2024;191 doi: 10.1016/j.lwt.2023.115640. [DOI] [Google Scholar]
  24. Tan C.S., Hassan M., Mohamed Hussein Z.A., Ismail I., Ho K.L., Ng C.L., Zainal Z. Structural and kinetic studies of a novel nerol dehydrogenase from Persicaria minor, a nerol-specific enzyme for citral biosynthesis. Plant Physiology and Biochemistry. 2018;123:359–368. doi: 10.1016/j.plaphy.2017.12.033. [DOI] [PubMed] [Google Scholar]
  25. Tian X., Wang H., Chen L., Yuan H., Peng C., Wang W. Distinct changes in metabolic profile and sensory quality with different varieties of Chrysanthemum (Juhua) tea measured by LC-MS-based untargeted metabolomics and electronic tongue [article] Foods. 2024;13 doi: 10.3390/foods13071080. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Wang B., Yu M., Tang Y., Wang Y., Xia T., Song H. Characterization of odor-active compounds in Dahongpao Wuyi rock tea (Camellia sinensis) by sensory-directed flavor analysis. Journal of Food Composition and Analysis. 2023;123 doi: 10.1016/j.jfca.2023.105612. [DOI] [Google Scholar]
  27. Wang H., Chen L., Xu A., Zhao Y., Wang Y., Liu Z., Xu P. Thermochemical reactions in tea drying shape the flavor of tea: A review. Food Research International. 2024;197 doi: 10.1016/j.foodres.2024.115188. [DOI] [PubMed] [Google Scholar]
  28. Wang Q., Qin D., Jiang X., Fang K., Li B., Wang Q.…Wu H. Characterization of the aroma profiles of Guangdong black teas using non-targeted metabolomics. Foods. 2023;12 doi: 10.3390/foods12071560. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Wang Y., Huang L., Deng G., Ning J. The shaking and standing processing improve the aroma quality of summer black tea. Food Chemistry. 2024;454 doi: 10.1016/j.foodchem.2024.139772. [DOI] [PubMed] [Google Scholar]
  30. Wang Z., Liang Y., Wu W., Gao C., Xiao C., Zhou Z., Lin F., Sun W. The effect of different drying temperatures on flavonoid glycosides in white tea: A targeted metabolomics, molecular docking, and simulated reaction study. Food Research International. 2024;190 doi: 10.1016/j.foodres.2024.114634. [DOI] [PubMed] [Google Scholar]
  31. Woodward K.A., Hopkins N.D., Draijer R., de Graaf Y., Low D.A., Thijssen D.H.J. Acute black tea consumption improves cutaneous vascular function in healthy middle-aged humans. Clinical Nutrition. 2018;37(1):242–249. doi: 10.1016/j.clnu.2016.12.013. [DOI] [PubMed] [Google Scholar]
  32. Wu S., Yu Q., Shen S., Shan X., Hua J., Zhu J., Qiu J., Deng Y., Zhou Q., Jiang Y., Yuan H., Li J. Non-targeted metabolomics and electronic tongue analysis reveal the effect of rolling time on the sensory quality and nonvolatile metabolites of congou black tea. LWT- Food Science and Technology. 2022;169 doi: 10.1016/j.lwt.2022.113971. [DOI] [Google Scholar]
  33. Xiao Y., He C., Chen Y., Ho C., Wu X., Huang Y., Gao Y., Hou A., Li Z., Wang Y., Liu Z. UPLC–QQQ–MS/MS-based widely targeted metabolomic analysis reveals the effect of solid-state fermentation with Eurotium cristatum on the dynamic changes in the metabolite profile of dark tea. Food Chemistry. 2022;378 doi: 10.1016/j.foodchem.2021.131999. [DOI] [PubMed] [Google Scholar]
  34. Xiao Y., Huang Y., Chen Y., Xiao L., Zhang X., Yang C., Li Z., Zhu M., Liu Z., Wang Y. Discrimination and characterization of the volatile profiles of five Fu brick teas from different manufacturing regions by using HS–SPME/GC–MS and HS–GC–IMS. Current Research in Food Science. 2022;5:1788–1807. doi: 10.1016/j.crfs.2022.09.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Xiao Z., Shen T., Niu Y., Ke Q., Sun Z., Yang E., Liu X., Zhang J., Zhu J. Unraveling the characteristic aroma compounds in Longjing tea and their interaction mechanisms with S-curve and broad-spectrum olfactory receptors using molecular docking. Food Bioscience. 2024;60 doi: 10.1016/j.fbio.2024.104423. [DOI] [Google Scholar]
  36. Ye J., Ye Y., Yin J., Jin J., Liang Y., Liu R., Tang P., Xu Y. Bitterness and astringency of tea leaves and products: Formation mechanism and reducing strategies. Trends in Food Science & Technology. 2022;123:130–143. doi: 10.1016/j.tifs.2022.02.031. [DOI] [Google Scholar]
  37. Yu P., Zhao X., Cai Y., Zhong N., Huang H., Gong Y., Chen Q., Zheng H. Variation in the taste quality and metabolites of black teas produced in early and late spring. Food Research International. 2025;216 doi: 10.1016/j.foodres.2025.116553. [DOI] [PubMed] [Google Scholar]
  38. Yu X., Chen X., Li Y., Li L. Effect of drying methods on volatile compounds of Citrus reticulata Ponkan and chachi peels as characterized by GC-MS and GC-IMS. foods. 2022;11:2662. doi: 10.3390/foods11172662. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Zeng S., Zhang L., Li P., Pu D., Fu Y., Zheng R., Xi H., Qiao K., Wang D., Sun B., Sun S., Zhang Y. Molecular mechanisms of caramel-like odorant-olfactory receptor interactions based on a computational chemistry approach. Food Research International. 2023;171 doi: 10.1016/j.foodres.2023.113063. [DOI] [PubMed] [Google Scholar]
  40. Zhai X.T., Zhang L., Granvogl M., Ho C.T., Wan X.C. Flavor of tea (Camellia sinensis): A review on odorants and analytical techniques. Comprehensive Reviews in Food Science and Food Safety. 2022;21(5):3867–3909. doi: 10.1111/1541-4337.12999. [DOI] [PubMed] [Google Scholar]
  41. Zhang H., Qi R., Mine Y. The impact of oolong and black tea polyphenols on human health. Food Bioscience. 2019;29:55–61. doi: 10.1016/j.fbio.2019.03.009. [DOI] [Google Scholar]
  42. Zhang J., Feng W., Xiong Z., Dong S., Sheng C., Wu Y.…Ning J. Investigation of the effect of over-fired drying on the taste and aroma of lu’an Guapian tea using metabolomics and sensory histology techniques. Food Chemistry. 2024;437 doi: 10.1016/j.foodchem.2023.137851. [DOI] [PubMed] [Google Scholar]
  43. Zhang L., Cao Q., Granato D., Xu Y., Ho C. Association between chemistry and taste of tea: A review. Trends in Food Science & Technology. 2020;101:139–149. doi: 10.1016/j.tifs.2020.05.015. [DOI] [Google Scholar]
  44. Zhu J., Liu X., Lin Y., Ke Q., Niu Y., Zhang J., Yang E., Shen T., Sun Z., Xiao Z. Unraveling the characteristic chestnut aroma compounds in MeiTanCuiYa green tea and their interaction mechanisms with broad-spectrum olfactory receptors using molecular docking. LWT- Food Science and Technology. 2024;194 doi: 10.1016/j.lwt.2024.115785. [DOI] [Google Scholar]
  45. Zhu J., Liu X., Sun Z., Shen T., Wang Z., Yang E.…Xiao Z. Reviewing the mechanisms of the interaction of characteristic aroma compounds of Longjing tea with S-curve, molecular docking, and molecular dynamics simulation: With a focus on (+)-(1R,2S)-methyl Epijasmonate and β-ionone as examples. Journal of Agricultural and Food Chemistry. 2025;73(6):3645–3656. doi: 10.1021/acs.jafc.4c08754. [DOI] [PubMed] [Google Scholar]
  46. Zhu Y., Jia H., Xi M., Li J., Yang L., Li X. Characterization of a naringinase from aspergillus oryzae 11250 and its application in the debitterization of orange juice. Process Biochemistry. 2017;62:114–121. doi: 10.1016/j.procbio.2017.07.012. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary material: Figure S1 Analysis of differential non-volatile metabolites of four samples. (A) Principal Component Analysis (PCA); (B) Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA); (C) results of 200 permutation test. FD, the freeze drying sample. FSD, first steaming and drying sample. SSD, second steaming and drying sample. TSD, third steaming and drying sample. Figure S2 Molecular docking simulation between 3-carene and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S3 Molecular docking simulation between caryophyllene and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S4 Molecular docking simulation between (E)-2-decenal and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S5 Molecular docking simulation between dodecanal and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S6 Molecular docking simulation between methyl salicylate and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S7 Molecular docking simulation between neral and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S8 Molecular docking simulation between nerol and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S9 Molecular docking simulation between neryl acetate and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S10 Molecular docking simulation between γ-terpinene and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S11 Molecular docking simulation between undecanal and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S12 Molecular docking simulation between α-terpineol and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Figure S13 Molecular docking simulation between β-thujene and OR1A1, OR1D2, OR1G1, OR2W1, OR52D1. Table S1 Electronic tongue test results of pomelo black tea. Table S2 Identification of key metabolites in samples through LC-MS (VIP > 1.0 and p < 0.05). Table S3 Total volatile compounds in four samples. Table S4 Volatile compounds with p < 0.05 in four samples. Table S5 Summary of the interaction forces between olfactory receptors and ligands.

mmc1.docx (3MB, docx)

Data Availability Statement

Data will be made available on request.

Articles from Food Chemistry: X are provided here courtesy of Elsevier

RESOURCES